#01 Orient
Why this, why now
The case for taking advanced AI seriously: capabilities, timelines, and what it would mean for things to go well or badly.
Swarthmore College · Fall 2026 · Applications open
Intro Fellowship · Fall 2026
Apply for Fall 2026Next session
SAISI-F26-FEL · The fellowship
An eight-week reading group on how powerful AI systems fail, who is working to prevent it, and what a Swarthmore student can actually do about it.
Eight weeks · In order
#01 Orient
The case for taking advanced AI seriously: capabilities, timelines, and what it would mean for things to go well or badly.
#02 Open
Pretraining, fine-tuning, RLHF, and scaling laws, explained without code. Enough to read the rest of the syllabus.
#03 Break
Specification gaming, reward hacking, goal misgeneralization. Real incidents, real transcripts.
#04 Doubt
Situational awareness, sandbagging, and whether we can keep a system that is smarter than its overseers in check.
#05 Look
Interpretability: what circuits, features, and probes have found so far, and what they have not.
#06 Test
Dangerous-capability evals, red teaming, and the organizations whose job is to say no before a launch.
#07 Govern
Compute governance, safety frameworks, export controls, and who actually decides.
#08 Act
Research programs, policy fellowships, career paths, and a short project you present to the cohort.
Organizers
Class of 2028. Runs the fellowship and the Kairos relationship.
Class of 2028. Facilitates a cohort, runs events and speaker outreach.
Class of 2028. Facilitates a cohort, runs reading groups and community.
Swarthmore faculty sponsor for the chartered organization.
Part of a network
SAISI is the first Tri-Co group in a network of student AI safety organizations, supported by Coefficient Giving through Project Kairos's Pathfinder fellowship.